How to increase performance on linux platform

Users Dgraph
1

I am using dgraph version 1.0.7 on linux (centos) (32 GB RAM).
I have downloaded dgraph-linux-amd64.tar from Releases · dgraph-io/dgraph · GitHub .

Started servers with below commands >>
dgraph zero
dgraph server --config config.json

My config.json file contains below data >>
{
“lru_mb”: “4096”,
“port_offset”: “9191”,
“zero”: “localhost:5080”,
“postings”: “/softwares/Dgraph/p”,
“wal”: “/softwares/Dgraph/w”,
“export”: “/softwares/Dgraph/exportDB”
}

I am able to use the dgraph and can mutate and query.
But it’s taking long time while mutating. what could be the possible reason?

while with the same config and setup, it’s very quick on windows 10 (16GB ram).

2

Can you share more information about your setup? For instance, are you using SSDs, how many machines are running Dgraph, etc.

3

No i am not using any SSDs.
and DGraph is running on single linux machine.
and That’s the only setup i have done dgraph.

4

Running a query on ratel query browser
{node (func: has(name)) {count(uid)}}
Here output is >> count = 200

While i am running below query >>
{node (func: has(name)) {uid name}}

Total time taken to give results
on windows >>
Server latency: 2ms
Rendering latency: 179ms

on linux >>
Server latency: 62ms
Rendering latency: 167ms

5

I have tried with dgraph v1.0.8 also, it’s same.

Linux binaries are critical for me. @dmai

6

We recommend using local SSDs with high IOPS for ideal performance for Dgraph.

7

But i am not using SSDs in windows also, but there i have seen better performance than in linux.

8

I have got one PC with SSD (120 GB) with 8 GB Ram, Linux OS and tried this Dgraph with my data.
this is giving better results than HDD in linux with 20 GB ram. but still not faster than windows (16 GB ram, HDD) .